Evolution of the Fatigue Failure Prediction Process from Experiment to Artificial Intelligence: A Review

Area of Science:

• Engineering
• Materials Science
• Computational Science

Background:

• Fatigue rupture prediction methods have evolved over time, with developmental stages becoming progressively shorter.
• Traditional methods face limitations due to numerous influencing factors and insufficient practical considerations.
• Recent advancements focus on integrating physics-based models with data-driven artificial intelligence (AI) approaches to enhance prediction accuracy.

Purpose of the Study:

• To present the evolutionary trajectory of fatigue rupture prediction methods.
• To highlight the shift towards hybrid approaches combining physical mechanisms with AI.
• To analyze the effectiveness of different review methodologies for this subject.

Main Methods:

• A combination of semi-systematic and integrative review methods was employed.
• These methods were chosen for their complementary strengths in analyzing the subject.
• The study analyzes the historical development and current trends in fatigue prediction.

Main Results:

• The experimental period represented the longest developmental stage in fatigue prediction evolution.
• The integration of advanced mathematical methods and AI has significantly accelerated the evolution and reduced implementation times.
• Despite advancements, all fatigue rupture prediction methods exhibit limitations.

Conclusions:

• The evolution of fatigue rupture prediction is characterized by decreasing developmental stages, driven by powerful mathematical and AI methods.
• Combining physics-based fatigue failure mechanisms with data-driven AI offers a promising avenue for increasing prediction accuracy.
• A hybrid review approach (semi-systematic and integrative) effectively complements the analysis of this evolving field.